Oncogene (2009) 28, 876–885 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE HAND1 expression is negatively regulated by the High Mobility Group A1 and is drastically reduced in human thyroid carcinomas

J Martinez Hoyos1,2, A Ferraro2, S Sacchetti2, S Keller1,2, I De Martino1,2, E Borbone2, P Pallante2, M Fedele1, D Montanaro2, F Esposito1, P Cserjesi3, L Chiariotti1,2, G Troncone4 and A Fusco1,2

1Dipartimento di Biologia e Patologia Cellulare e Molecolare e/o Istituto di Endocrinologia ed Oncologia Sperimentale del CNR, Facolta` di Medicina e Chirurgia di Napoli, Universita` degli Studi di Napoli ‘Federico II’, Naples, Italy; 2NOGEC (Naples Oncogenomic Center), CEINGE Biotecnologie Avanzate, Naples, Italy; 3Department of Cell and Molecular Biology, Tulane University, New Orleans, LA, USA and 4Dipartimento di Anatomia Patologica e Citopatologia, Facolta` di Medicina e Chirurgia, Universita` di Napoli ‘Federico II’, Naples, Italy

HMGA1 proteins exert their major physiological function Introduction during embryonic development and play a critical role in neoplastictransformation. Here, we show that Hand1 The high mobility group A (HMGA) family gene, which codes for a crucial for includes HMGA1a and HMGA1b, which are encoded differentiation of trophoblast giant cells and heart by the same gene through alternative splicing, and the development, is upregulated in minus embryonic closely related HMGA2 protein (Fedele et al., 2001; stem cells. We demonstrate that HMGA1 proteins bind Fusco and Fedele, 2007). These proteins are non-histone directly to Hand1 promoter both in vitro and in vivo and architectural nuclear factors that bind the minor groove inhibit Hand1 promoter activity. We have also investi- of AT-rich DNA sequences through three short basic gated HAND1 expression in human thyroid carcinoma repeats, called ‘AT-hooks’, located at the NH2-terminal cell lines and tissues, in which HMGA proteins are region of the proteins (Thanos and Maniatis, 1995; overexpressed, with respect to normal thyroid; an inverse Reeves, 2001). The involvement of HMGA proteins in correlation between HMGA1 and HAND1 expression embryogenesis, cell proliferation, differentiation, apop- was found in all thyroid tumor histotypes. A correlation tosis and above all, cancer development has been between HAND1 gene repression and promoter hyper- extensively demonstrated (Fedele et al., 2001; Reeves, methylation was found in anaplastic carcinomas but not in 2001; Fusco and Fedele, 2007). In particular, HMGA other thyroid tumor histotypes. Therefore, we can proteins seem to play their major physiological role hypothesize that HMGA1 overexpression plays a key during embryonic development; in fact, their expression role on HAND1 silencing in differentiated thyroid is very high during embryogenesis, whereas it is very low carcinomas and that promoter hypermethylation occurs or negligible in normal adult tissues (Chiappetta et al., in later stages of thyroid tumor progression. Finally, the 1996). HMGA1 proteins have been found abundant in restoration of the HAND1 reduces the several malignant neoplasias, including colorectal, clonogenic ability of two human thyroid carcinoma- prostate, cervical, lung and thyroid carcinoma (Fedele derived cell lines, suggesting that HAND1 downregulation et al., 2001; Fusco and Fedele, 2007) and, recently, also may have a role in the process of thyroid carcinogenesis. in glioblastoma (Donato et al., 2004). Oncogene (2009) 28, 876–885; doi:10.1038/onc.2008.438; We identified HMGA1-regulated analysing by published online 8 December 2008 microarrays the expression profile of murine embryonic stem (ES) cells carrying two, one and no hmga1 Keywords: HMGA1; HAND1; knockout mice; ES cells; functional alleles (Martinez Hoyos et al., 2004; Fedele thyroid carcinomas et al., 2006). In this study, we focused our attention on the Hand1 gene, which showed by microarrays a 14-fold change in the hmga1-null ES in comparison to the wild-type cells (Martinez Hoyos et al., 2004). Hand1 (also named eHand/Hxt/Thing1) belongs to the Twist subfamily of Class B bHLH transcription factors. Murine Hand1 is expressed in developing heart tissue and derivatives of neural crest cells (Cserjesi et al., 1995). Extra-embryonic mRNA production has been Correspondence: Dr A Fusco, Dipartimento di Biologia e Patologia detected in the ectoplacental cone, in giant cells as well Cellulare e Molecolare, Facolta` di Medicina e Chirurgia di Napoli, as in distinct regions of the spongiotrophoblast cell layer via Pansini 5, 80131 Naples, Italy. (Cross et al., 1995). Embryos carrying a homozygous E-mail: [email protected] Received 21 May 2008; revised 16 October 2008; accepted 1 November mutation of Hand1, arrested at E7.5 with defects in 2008; published online 8 December 2008 trophoblast giant cells differentiation. Upon rescue of HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 877 the placental defect by aggregating wild-type tetraploid ES cells ES cells +/+ +/– –/– +/+ +/– –/– embryos, the fetuses died at E10.5 due to cardiac failure, 1 demonstrating that Hand1 has essential roles in both Hand1 0.5 trophoblast giant cell differentiation and cardiac mor- Hand1 0 phogenesis (Riley et al., 1998). In humans, HAND1 Hand2 HMGA1 -0.5 expression has been detected in trophoblast-like cells, log10 RQ the amniotic epithelium and in adult heart tissue, HMGA1 -1 suggesting that the protein may fulfill similar functions β-actin (Knofler et al., 2002). The mechanisms, however, which initiate and maintain HAND1 expression in extra- heart thyroid embryonic cell types are not well known. A role of +/+ +/– –/– +/+ +/– –/– HAND1 in cell proliferation and neoplastic transforma- tion has been recently envisaged because HAND1 gene Hand1 has been found silenced and hypermethylated in human β-actin gastric (Kaneda et al., 2002), pancreatic (Hagihara et al., 2004) and ovarian carcinomas (Takada et al., 2004). Figure 1 Hand1 expression in hmga1-knockout cells and tissues. Here, we demonstrate that Hand1 is upregulated in Semiquantitative (a and c) and quantitative (b) RT–PCR analyses hmga1-null ES cells, and that HMGA1 proteins bind for Hand1 and Hand2 expression were performed on RNA directly to the Hand1 promoter in vitro and in vivo extracted from wild-type ( þ / þ ), hmga1-single knockout ( þ /À) and hmga1-double knockout (À/À) ES cells and tissues. b-actin resulting in the inhibition of its activity. We also found expression was evaluated as internal control. that HAND1 was downregulated in human thyroid tumors. Such downregulation was associated with when we analysed the same tissues from the -null HMGA1 overexpression and, limited to anaplastic mice, no changes in Hand1 expression were observed carcinomas, with promoter hypermethylation. The (data not shown), indicating that Hand1 regulation was restoration of HAND1 expression led to a reduced cell HMGA1-specific. growth of two human papillary thyroid carcinoma cell It is noteworthy that no changes in the expression of lines, indicating a critical role of the HAND1 down- Hand2, another gene that belongs to the Twist subfamily regulation, likely mediated by HMGA1, in thyroid of class B bHLH transcription factors, were observed carcinogenesis. neither in ES cells, MEFs, nor in adult tissues from hmga1-andhmga2-knockout mice (Figures 1a and c).

Results HMGA1 proteins bind to murine and human Hand1 promoter in vitro and in vivo Loss of HMGA1 correlates with an increased Hand1 To evaluate whether the differential Hand1 gene expression in murine ES cells,heart and thyroid tissues expression was a direct effect of HMGA1 binding to Expression profile, as assessed by microarray analysis Hand1 gene regulatory regions, we performed an of ES cells bearing one or two disrupted hmga1 alleles, electrophoretic mobility shift assay. In particular, we identified Hand1 as a candidate gene negatively regu- analysed a region spanning nucleotides from À2326 to lated by HMGA1. In fact, microarray data showed an À2296 related to the transcription start site (TSS) of the increase of 3.8- and a 14.0-fold for the heterozygous and murine Hand1 gene which contains AT-rich putative homozygous ES cells, respectively, when compared with HMGA1 binding sites. As shown in Figure 2a, a wild-type cells (Martinez Hoyos et al., 2004). Our first recombinant HMGA1 protein was able to bind directly aim was to validate the results obtained by microarray to this region. Binding specificity was demonstrated by analysis by semiquantitative and quantitative RT–PCR competition experiments showing loss of binding with (Figures 1a and b). These analyses confirmed that the addition of 200-fold molar excess of specific, Hand1 was strongly overexpressed in hmga1-knockout unlabeled oligonucleotides. Subsequently, we performed ES compared with wild-type cells. Data clearly showed binding assays with total extracts from wild-type and that regulation of Hand1 expression was HMGA1-dose- hmga1-knockout ES cells. A specific complex, with a dependent as an intermediate level of Hand1 expression mobility corresponding to the HMGA1 proteins, was was observed in the hmga1-heterozygous ES cells. present in extracts from wild-type mice whereas it was The analysis of Hand1 expression in heart and thyroid absent in extracts from homozygous hmga1-knockout tissues derived from hmga1-knockout mice also revealed ES cells. This complex was specifically displaced a negative regulation by HMGA1 proteins (Figure 1c). by incubation with HMGA1-specific antibodies Conversely, no changes in Hand1 expression were (Figure 2b), demonstrating the specificity of the observed when embryonic fibroblasts, brain, spleen, HMGA1 binding to the Hand1 promoter. pancreas, liver, kidney and thymus from hmga1-null To verify whether HMGA1 proteins bind to murine mice were analysed (data not shown). These results are and human Hand1 promoter in vivo we performed consistent with the general concept that HMGA1- chromatin immunoprecipitation (ChIP) experiments. mediated gene regulation depends on the cellular Anti-HMGA1 antibodies precipitated murine Hand1 context (Martinez Hoyos et al., 2004). Interestingly, promoter from wild-type and heterozygous, but not

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 878 5 ng HMGA1 + – + – αHMGA1 ––+ –– 20 ng HMGA1 – + – – 200x oligo – – + – 200x oligo – ++––

HMGA1 Free probe

Free probe

Hmga1+/+ Hmga1–/–

IP HMGA1IP HA INPUT INPUT IgG IP HMGA1

+/++/– –/–– +/+ +/+ +/– /– HMGA1 – + – + – +

Hand1 prA

Hand1 prB IP HMGA1 IP HA INPUT +/+ –/––+/+/–– +/– /–

GAPDH pr

Figure 2 HMGA1 binding to Hand1 upstream regulating region. (a) Electrophoretic mobility shift assay performed incubating radiolabeled oligonucleotides spanning from À2326 to À2296 of the 50 untranslated region of the murine Hand1 gene with 5 or 20 ng of the recombinant HMGA1 as indicated. To assess the specificity of the binding, a 200 Â molar excess of unlabeled oligonucleotides were incubated as specific competitor. (b) Electrophoretic mobility shift assay performed with the same oligonucleotides as in (a), incubated with total extracts from wild-type ( þ / þ ) and hmga1- knockout (À/À) ES cells. Where indicated, the extracts were pre- incubated with HMGA1-specific antibodies. (c) Chromatin immunoprecipitation (ChIP) assay was performed on wild-type, hmga1- single knockout ( þ /À) and hmga1-double knockout (À/À) ES cells. The presence of À2367 to À2246 sequence of the murine Hand1 promoter was detected by PCR. Anti-HA was used as a negative immunoprecipitation control. (d) ChIP assay was performed on wild-type and hmga1-double knockout (À/À) thyroid tissue. Murine Hand1 promoter region À2367 to À2246 was detected by PCR. Anti-HA was used as a negative immunoprecipitation control. (e) Soluble chromatin from human HEK293 cells untransfected or transfected with HMGA1 was immunoprecipitated with anti-HMGA1 antibodies. The DNAs were then amplified by semiquantitative PCR using primers that cover two regions of human HAND1 promoter region (À1589/À1294 and À1311/À1020). Coimmunoprecipita- tions with isotype-matched IgG were also performed as control of the specificity of the interactions. To assess the specificity of the binding, the panel shows PCR amplification of the immunoprecipitated DNA using primers for the human GAPDH gene promoter.

from homozygous hmga1-knockout ES cells (Figure 2c). luciferase gene under the control of the mouse Hand1 No amplification was observed in samples immunopre- promoter region from nucleotides À2367 to À2084 cipitated with an unrelated antibody. Analogous results related to TSS. As shown in Figure 3a, when the were obtained when the ChIP experiments were HMGA1 expression vector was co-transfected, a reduc- performed on thyroid tissues originating from wild-type tion of the luciferase activity was observed in a dose- and hmga1-null mice (Figure 2d). dependent manner. No decrease in Hand1 promoter Occupancy of human HAND1 promoter region by activity was obtained when the cells were co-transfected HMGA1 was detected in anti-HMGA1-precipitated with a construct expressing HMGA2. Then, we gener- chromatin from HEK293 (Figure 2e). As HMGA1 ated two point mutations in the putative binding site for proteins are lowly expressed in HEK293, a weak signal the HMGA1 protein, replacing adenine À2320 and was detected in untransfected cells whereas a stronger thymidine À2319 with two guanines. Overexpression of signal was detected in HEK293 transfected with HMGA1 was able to reduce the activity of the wild-type HMGA1. No amplification was observed in samples construct, but it completely failed in its inhibitory effect immunoprecipitated with an unrelated antibody. if the HMGA1-binding site was mutated (Figure 3b).

HMGA1 proteins repress murine Hand1 promoter Hand1 expression is drastically decreased in thyroid cell To investigate the functional effect of HMGA1 binding transformation to Hand1 promoter, we transiently transfected the rat The HMGA1 proteins are overexpressed in several thyroid cells FRTL-5, that do not express detectable malignant neoplasias. To evaluate a possible role of levels of HMGA1, with a construct expressing the Hand1 regulation by HMGA1 in cell transformation, we

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 879 2.5

2 3.5 3 1.5 2.5 2 1 1.5

Fold of activation 0.5 1

Fold of activation 0.5 0 0 HMGA1 (µg) – 0.5 1 2 4 – HMGA1(µg) – 2 – – HMGA2 (µg) –– –––4 Figure 3 HMGA1 proteins repress murine Hand1 promoter. (a) Effect of HMGA1 expression on the activity of murine Hand1 promoter transfected in FRTL-5 cells. (b) Effect of HMGA1 expression on the activity of murine Hand1 promoter (black boxes) and same promoter in which A À2320 and T À2319 were replaced with G (gray boxes). The results are reported as mean expression values of three independent experiments, with bars indicating standard deviation (mean±s.d.).

analysed, by RT–PCR, the expression of Hand1 in thyroid areas were scored for the expression of HAND1 FRTL-5 normal rat thyroid cells, which do not express by quantitative analysis performed with a computerized significant levels of HMGA1, in the same cells analyzer system (Ibas 2000, Kontron, Zeiss) and relative transformed by the Kirsten murine sarcoma virus data were reported in Table 1. As shown in Figure 5 (KiMSV), which express high levels of HMGA1 (Panels a and b), most of the normal thyroid follicular (FRTL-5 –KiMSV) and in KiMSV-infected FRTL-5 cells expressed the HAND1 protein at a very high level cells in which the expression of HMGA1 was down- (median percentage of stained cells 66%) with the regulated by an antisense construct (FRTL-5- staining evenly distributed in all follicles. A high HMGA1as-KiMSV cells). A significant Hand1 expres- percentage of HAND1 nuclear labeling was also sion was observed in the cells, which do not express the displayed by FTA (67% median value), whereas HMGA1 proteins, such as the FRTL-5 and FRTL-5- HAND1 expression was reduced in both differentiated HMGA1as-KiMSV, whereas no expression was and undifferentiated malignant histotypes. In the former observed in the FRTL-KiMSV cells expressing high group, HAND1 median expression values were 22% in HMGA1 levels (Figure 4a). These results confirm an PTCand 25% in FTC,whereas in ATC HAND1 loss of inverse correlation between HMGA1 and Hand1 expression was even more evident (15% median value). expression also in transformed cells. In Figure 5, (at low and high magnification, respec- Subsequently, we analysed the expression of HMGA1 tively) the sporadic HAND1 staining of scattered and HAND1 in human thyroid carcinoma cell lines and neoplastic cells typical of PTC(Panels c and d), FTC tissues in comparison with normal controls. To this aim (Panels e and f) and ATC(g and h) is shown. No a panel of 12 human thyroid carcinoma-derived cell lines staining was observed when normal thyroid gland and 20 surgically removed human thyroid carcinomas samples were stained with antibodies pre-incubated with were analysed by semi-quantitative and quantitative the peptide against which the antibodies were raised RT–PCR, respectively (Figures 4b and c). Again, an (Figure 5, Panels i and l) or in the absence of the primary inverse correlation between HMGA1 and HAND1 antibodies (data not shown). Therefore, immunohisto- expression levels was observed. In fact, HMGA1 chemical analysis confirmed that HAND1 is expressed expression was increased in all the tumor-derived cell in normal thyroid and in benign neoplastic lesions and lines and tumors tested when compared with normal that its expression is significantly decreased in differ- primary cultured cells and normal tissues, respectively. entiated and undifferentiated carcinomas (Po0.001) Conversely, HAND1 expression was much lower in all (Figure 6). the thyroid carcinoma cell lines and tumor samples than in normal cells and tissues, respectively (Figures 4b and c). LOH and DNA methylation analyses of HAND1 gene To find out the mechanisms, other than HMGA1 Immunohistochemical analysis of HAND1 expression overexpression, underlying the block of HAND1 in human thyroid carcinomas expression in thyroid carcinomas, we analysed 37 Subsequently, we carried out an immunohistochemical informative carcinoma samples including 30 PTCs and analysis of paraffin-embedded thyroid tissues. Ten cases seven ATCs for loss of heterozygosity. The analysis of of follicular thyroid adenoma (FTA), 15 cases of eight single nucleotide polymorphisms showed that all classical papillary thyroid carcinoma (PTC), 15 cases the carcinoma samples were negative for loss of of follicular thyroid carcinoma (FTC), 10 cases of heterozygosity. Then, we investigated whether epige- anaplastic thyroid carcinoma (ATC) and 10 normal netic mechanisms may contribute to silence HAND1

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 880

as- ab

KiMSV HMGA1

FRTL-5 FRTL-5- FRTL-5- KiMSV

Hand1 cd HMGA1

β-actin TPC1 ARO FRO FB-1 KAT-4 TN WRO FB-2 BC-PAP NIM N-PAP KAT-18 WRO/ARO ef HAND1

HMGA1

β-actin

gh FTC MTC PTC ATC 1

0.5

0 HMGA1 Hand1

log10 RQ il -0.5

-1 1 2 3 4 5 6 7 8 9 10 1112 13 14 15 1617 18 19 20 21 Figure 4 Hand1 expression is decreased in thyroid tumorigenesis. (a) RT–PCR analysis of Hand1 and HMGA1 expression in rat thyroid cells. Sources of RNA are FRTL-5, rat thyroid epithelial cell line; FRTL-5-KiMSV, FRTL-5 infected with the Kirsten Figure 5 Immunohistochemical analysis of HAND1 protein murine sarcoma virus; FRTL-5-HMGA1as-KiMSV, FRTL-5 expression in benign and malignant thyroid tissues. Paraffin transfected with a construct carrying HMGA1 mRNA in an sections from hyperlastic and neoplasic thyroid tissues were antisense orientation, and then infected with the Kirsten murine analysed by immunohistochemistry using antibodies raised against sarcoma virus. (b) RT–PCR analysis of HAND1 and HMGA1 a specific HAND1 peptide. (a and b) Immunostaining of a normal expression in human thyroid carcinoma-derived cell lines. thyroid (  10 and  40 magnification, respectively). A strong (c) Quantitative RT–PCR analysis was performed on human positive reactivity was observed. (c and d) Immunostaining of a thyroid samples of different histotype. Sources of RNA are: 1, thyroid papillary carcinoma (  10 and  40 magnification, normal thyroid (NT); 2, follicular thyroid carcinoma (FTC); 3, respectively). The signal is weak or absent in the neoplastic cells. medullary thyroid carcinoma (MTC); 4–14, papillary thyroid (e and f) Immunostaining of a thyroid follicular carcinoma (  10 carcinoma (PTC); 15–21, anaplastic thyroid carcinoma (ATC). and  40 magnification, respectively). The signal is weak or absent in the neoplastic cells. (g and h) Immunostaining of an anaplastic thyroid carcinoma (  10 and  40 magnification, respectively). No positivity was observed. (i and l) Immunostaining of a normal thyroid with antibodies pre-incubated with the peptide against Table 1 HAND1 protein immunohistochemical expression in which the antibodies were raised (  10 and  40 magnification, different diagnostic categories respectively). No immunoreactivity was detected. Diagnostic Category N Median Range Min Max

Non-neoplastic thyroid 10 66 21 57 78 Thyroid adenoma 10 67 33 45 78 À109) that have been previously shown to be hyper- Papillary thyroid carcinoma 15 22 27 11 38 Follicular thyroid carcinoma 15 25 31 14 45 methylated in gastric, pancreatic and ovarian tumors Anaplastic thyroid carcinoma 10 15 15 11 26 (Kaneda et al., 2002; Hagihara et al., 2004 and Takada et al., 2004). A representative pyrogram showing a bisulfite methylation profile of these CpG sites is shown gene in thyroid tumors. We performed a quantitative in Figure 7. Comparison of HAND1 methylation state DNA methylation analysis of HAND1 promoter using (Figure 7c) with HAND1 expression levels (Figure 4c) in pyrosequencing technology. We chose to analyse four thyroid tumor samples, indicated that a significant CpG sites (nucleotide positions À85, À94, À96 and correlation between HAND1 hypermethylation and

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 881 lower number of colonies compared with the cells transfected with the empty vector. This result suggests that HAND1 downregulation could have an important role in thyroid carcinoma cell proliferation, and then in the process of carcinogenesis.

Discussion

In our previous work, we identified HMGA1-regulated genes analysing by microarrays the expression profile of embryonic stem (ES) cells bearing one or two disrupted hmga1 alleles (Martinez Hoyos et al., 2004). The generation of hmga1 null mice unveiled a critical role of HMGA1 proteins on cardiomyocytic cell growth: in fact, both heterozygous and homozygous mice for the hmga1-null allele showed cardiac hypertrophy. Moreover, these mice also developed hemato- logic malignancies, including B-cell lymphoma and myeloid granuloerythroblastic leukemia (Fedele et al., 2006). Here, we focused our attention on Hand1, a gene that encodes a transcription factor crucial for differentiation of trophoblast giant cells and heart development (Riley et al., 1998). We validated the microarray results by semiquantitative and quantitative RT–PCR in ES cells. As in microarrays, we found that Hand1 expression displayed HMGA1 dose-dependency: the phenotype of heterozygous cells was intermediate between those of wild-type and homozygous knockout cells. We found an increased Hand1 expression also in heart and thyroid but not other tissues from hmga1-knockout mice with respect to wild-types. These results indicated that Figure 6 Boxplot and results of Kruskal–Wallis test for HAND1 protein expression in different diagnostic categories. Data are HMGA1-mediated Hand1 regulation depends on the displayed as median, highest and lowest values and an interquartile cellular context, as occurs for other HMGA1-regulated range containing 50% of values. genes (Martinez Hoyos et al., 2004). In this report, we show that HMGA1 proteins bind directly to Hand1 promoter in vitro and in vivo,by electrophoretic mobility shift assay and chromatin silencing is present in anaplastic carcinomas. By immunoprecipitation experiments. Moreover, this bind- contrast, no such correlation was observed in the other ing has a functional effect as we have seen by luciferase thyroid carcinoma histotypes. These data suggest that assay that hmga1 expression reduces the activity of silencing of HAND1 in papillary carcinomas is primar- Hand1 promoter. When we analysed by RT–PCR the ily because of HMGA1 overexpression whereas, in the expression of Hand1 in MEF and adult tissues taken most aggressive and undifferentiated ATC, additional from hmga2-knockout mice, we did not find changes mechanisms of epigenetic nature may contribute to keep between wild-type and knockout cells and tissues. This HAND1 repressed. result indicated that Hand1 was probably not regulated by HMGA2. In fact, when we co-transfected hmga2 expression construct together with Hand1 promoter we Restoration of HAND1 gene expression inhibits did not find any alteration in Hand1-promoter activity the growth of thyroid carcinoma cell lines by luciferase assay. This specific responsiveness to The reduced expression of HAND1 in thyroid carcino- HMGA1 and not HMGA2 confirms that even though mas prompted us to investigate if its downregulation HMGA1 and HMGA2 have a similar structure and could be a causative event of the thyroid cell prolifera- expression profile (high during embryogenesis and in tion. To this purpose we performed a colony-forming neoplastic tissue), they exert different functions. This is assay on the NPA and FB2 human thyroid carcinoma consistent with our previous findings showing that the cells transfecting them with 5 and 10 mg of pCDNA3- phenotype of the hmga1- and hmga2-knockout mice is Hand1 or the corresponding empty vector. As shown in divergent: cardiac hypertrophy and B-cell lymphomas in Table 2, the cells transfected with the HAND1 expres- hmga1-null mice and a reduction in size and fat tissue sion vector gave rise to a significant dose-dependent in hmga2-null mice (Fedele et al., 2006).

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 882

Figure 7 DNA methylation analysis on HAND1 gene. (a) Sequence of the analysed HAND1 promoter region. The transcriptional start site is underlined. The analysed CpGs are boxed. (b) Representative pyrogram. The four targeted cytosines are enclosed in unshaded squares (as reverse strand was read, G peaks—arrowed—indicate methylated cytosine whereas A indicates unmethylated cytosine). The control, non-CpG cytosine residue showing complete conversion of cytosine to uracil by bisulfite treatment, is shown. (c) Hystogram of pyrosequencing results. For each sample the percentage of methylation represents the average of the methylation degree of the four CpG sites analysed. MTC, FTC, PTC and ATC indicate medullary, follicular, papillary and anaplatic thyroid carcinomas, respectively. NT, Normal thyroid samples.

Table 2 Restoration of HAND1 gene expression reduces clonogenic Hand2. This gene was on the microarray but was not capacities of thyroid cancer cells changed in the hmga1-knockout cells compared with Cell lines Plasmids Number of clones wild-types. Anyway, we performed semiquantitative PCR, but we did not find altered expression of Hand2 FB-2 5 mg empty vector 95±11.7 neither in ES cells, MEFs, nor in adult tissues from FB-2 5 mg Hand1 75±10* hmga1- and hmga2-knockout mice. FB-2 10 mg empty vector 99±9.5 FB-2 10 mg Hand1 58±12.2* HAND-1 and -2 gene expression has been studied in N-PA 5 mg empty vector 258±13.3 normal and hypertrophied human ventricle (Ritter et al., N-PA 5 mg Hand1 167±16.7* 1999) and found increased in both the hypertrophied left N-PA 10 mg empty vector 263±11.4 and right ventricle of patients with hypertrophic ± N-PA 10 mg Hand1 122 14.6* obstructive cardiomyopathy and tetralogy of Fallot. The results are reported as mean expression values with bars indicating Consistently, we found an increased Hand1 mRNA standard deviation (mean±s.d.). expression in the hypertrophied heart of hmga1-knock- *Po0.05 with respect to the negative control (empty vector). out mice. Given this parallelism, it would be interesting to see the status of HMGA1 proteins in those patients. As Hand1 shares 87% homology in bHLH region Therefore, it is reasonable to hypothesize that the with Hand2, another gene that belongs to the Twist HMGA1-mediated regulation of Hand1 may have a subfamily of class B bHLH transcription factors, we critical role in the cardiac hypertrophy that develops in were intrigued if HMGA proteins could regulate even the absence of an appropriate HMGA1 expression.

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 883 The other tissue in the hmga1-knockout mice where USA) KpnI–XhoI cloning site. The point mutations in the we found Hand1 upregulation was the thyroid. In our HMGA binding site of the Hand1 promoter were generated laboratory, we have previously demonstrated that using the QuikChange Mutagenesis Kit (Stratagene, La Jolla, CA, USA) in accordance with the manufacturer’s protocols. HMGA1 proteins have a major role in thyroid 0 transformation (Berlingieri et al., 2002). As, in the The primers used were: 5 -ggcaggaggagactctgttattaCCtaattagt taaaata-30 (forward) and 50-tattttaactaattaGGtaataacagagtctc present paper, we demonstrate that HMGA1 proteins ctcctgcc-30 (reverse), where point mutations are shown in regulate Hand1, we suspected that Hand1 could have a uppercase type. Hand1, HMGA1 and HMGA2 expression role in thyroid cell transformation. First of all, we plasmids were constructed by cloning the murine full-length studied Hand1 expression in a cell system constituted by cDNAs of Hand1, Hmga1b or Hmga2 into the mammalian normal rat thyroid cells (FRTL-5) that do not express expression vector pcDNA3.1 (Invitrogen). HMGA1 proteins, the same cells malignantly trans- formed by the KiMSV (FRTL-5-KiMSV) that express Cell culture and transfections high HMGA1 levels and, FRTL-5-KiMSV cells in The generation and culture of hmga1 þ /À and hmga1À/À ES which the synthesis of the HMGA1 protein was blocked cells are described elsewhere (Battista et al., 2003). FRTL-5, by an antisense construct (FRTL-5-HMGA1as- FRTL-5-KiMSV and FRTL-5–HMGA1as-KiMSV cells and KiMSV). We found Hand1 expression in cells that do their culture conditions, are reported elsewhere (Berlingieri not express HMGA1 proteins but not in cells expressing et al., 2002). Cells were harvested 24 h post-transfection and lysates were analysed for luciferase activity. Transfection HMGA1 proteins, confirming an inverse correlation efficiency was normalized using the b-galactosidase activity between HMGA1 and Hand1 expression. When we and fold of activation were calculated by dividing for pGL3 analysed by RT–PCR 12 human thyroid carcinoma- luciferase activity. All the assays were performed in triplicate derived cell lines and by real-time–PCR 20 surgically and repeated in three independent experiments. removed human thyroid carcinomas, we found again an Human thyroid primary cultures and carcinoma cell lines inverse correlation between HMGA1 and HAND1 (TPC-1, WRO, NPA, ARO, FRO, NIM 1, B-CPAP, FB-1, expression. FB-2, Kat-4 and Kat-18) are described elsewhere (Pallante As it has been described that HAND1 is silenced and et al., 2005). For cloning efficiency assays, FB-2 and NPA cells hypermethylated in gastric, pancreatic and ovarian were either transfected with 5 and 10 mg of pCDNA3-Hand1 or carcinomas (Kaneda et al., 2002; Hagihara et al., 2004; the corresponding empty vector. Then, 48 h post-transfection of the plasmids 5 105 cells were plated on 10-cm dishes and Takada et al., 2004), we analysed the methylation status  cultured with a medium supplemented with G418 (Invitrogen) of HAND1 promoter in thyroid carcinomas of different for selection of transfected cells. Colonies were stained with histotypes and compared with normal thyroid. A 1.5% glutaraldehyde and 0.06% methylene blue in Hanks’ significant correlation between HAND1 hypermethyla- balanced salt solution and counted. tion and gene repression was found only in anaplastic All the transfections were performed by Lipofectamine 2000 carcinomas whereas only in few cases of papillary (Invitrogen), as suggested by the manufacturer. carcinomas HAND1 silencing was accompanied by a high methylation degree of the promoter region. Based Tissue samples on the knowledge that ATCs derive from PTCs by Neoplastic human thyroid tissues and normal adjacent tissue neoplastic progression (Kondo et al., 2006), we suggest or the controlateral normal thyroid lobe were obtained from that HMGA1 may have a critical role in HAND1 surgical specimens and immediately frozen in liquid nitrogen. silencing in PTCs, whereas in later stages of thyroid Thyroid tumors were collected at the Service d’Anatomo- tumor progression, promoter hypermethylation could Pathologie, Centre Hospitalier Lyon Sud, Pierre Benite, France. stabilize HAND1 repression. In conclusion, our data clearly demonstrate that HMGA1 proteins are able to directly downregulate RNA extraction from tissues and cells Total RNAs were extracted from frozen tissues and cell culture HAND1 expression in embryonic and adult tissues, and using TRI REAGENT (Molecular Research Center Inc., also in cell transformation. Moreover, the reduction of Cincinnati, OH, USA) solution, according to the manufac- the clonogenic ability after the restoration of the turer’s instructions. HAND1 gene expression in two human thyroid carci- noma-derived cell lines suggests that HAND1 down- Semiquantitative and quantitative RT–PCR regulation by HMGA1 proteins may have a role in the RNAs were treated with DnaseI (Invitrogen) and reverse- process of thyroid carcinogenesis. transcribed using random exonucleotides as primers and MuLV reverse transcriptase (Applied Biosystems, Foster City, CA, USA). For semiquantitative PCR, reactions were optimized for the number of cycles to ensure product intensity Materials and methods within the linear phase of amplification. Digitized data were analyzed using Imagequant (Molecular Dynamics, Sunnyvale, Plasmids CA, USA). Quantitative PCR was performed in triplicate with For the Hand1 promoter construct, the region from À2367 to SYBR Green PCR Master Mix (Applied Biosystems). Primers À2084, related to transcription start sites (TSS), of the mouse for amplification of Hand1 gene were: 50-gatgccttctcgagttaaaa- Hand1 gene was amplified using as primers 50-ctgagatcccagat 30 (forward) and 50-aagtgtagcgacaagaagga-30 (reverse), for cactca-30 (forward) and 50-gggatacacgaaggtcagtttt -30 (reverse), murine; 50-gttcaggacccaaaaagg-30 (forward) and 50-gcagagtctt- cloned in TA Cloning Vector (Invitrogen, Singapore, gatcttggag-30 (reverse), for rat; 50-ctggctctttctctcttgtc-30 (for- Singapore) and subcloned in pGL3 (Promega, Madison, WI, ward) and 50-cgtctggttctctttctcag-30 (reverse), for human.

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 884 Primers for amplification of Hand2 gene were: 50-ccaaactctc- provides an internal control of the completeness of bisulphite caagatcaag-30 (forward) and 50-ctcttctcctctttcacgtc-30 (reverse). treatment.

Electrophoretic mobility shift assay SNP-based Loss of Heterozygosity analysis Five or 20 ng of recombinant protein (Baldassarre et al., 2001) The primers, SNP reference (NCBI database) and alleles were: were incubated with radiolabeled double-strand oligonucleo- 50-cgaaataggcaaacaggctc-30 (forward) and 50-aaagctcatccagg tides, corresponding to the region spanning bases from -2326 gacga-30 (reverse) for rs924581 (A/G); 50-gaagacccgatctgttt to -2296 with respect of TSS of the murine Hand1 gene tacct-30 (forward) and 50-cttcaaggctgaactcaagaa-30 (reverse) (50-ttattaattaattagttaaaataaaataaat-30). A 200-fold excess of for rs4370323 (A/G), rs1846966 (C/G), rs11748765 (A/T); specific unlabeled competitor oligonucleotide was added. The 50-cgctgttaatgctctcagt-30 (forward) and 50-gtaaaacctgggatagc- same oligonucleotides were also used in binding assays with ca-30 (reverse) for rs6880185 (A/G), rs13171812 (C/T), total extract from wild-type and hmga1-knockout ES cells. rs993098 (A/C) and rs3822714 (A/G). PCR conditions were: Experimental conditions and anti-HMGA1 antibodies used 94 1Cfor 2 min, 35 cycles at 94 1Cfor 20 s, 58 1Cfor 10 s, 70 1C were as previously described (Martinez Hoyos et al., 2004). for 40 s followed by 5 min at 70 1C.

Chromatin immunoprecipitation Immunohistochemical analysis Chromatin immunoprecipitation was performed as previously Immunostaining was performed with indirect-peroxidase described (Martinez Hoyos et al., 2004). For mouse Hand1 assay, as previously described (Pallante et al. 2005). Archivial promoter, the primers used were: 50-ctgagatcccagatcactca-30 tumor blocks were from ‘Dipartimento di Scienze Biomorfo- (forward) and 50-cttggtgacaagcacctt-30 (reverse). For human logiche e Funzionali’ (University of Naples, Naples, Italy). HAND1 promoter were analyzed two sequences spanning A polyclonal antibody specific for the carboxy-terminal region nucleotides from -1589 to -1294 and from -1311 to -1020 of human HAND1 (1:200; Sigma, St Louis, MO, USA) was related to TSS. The primers used were: 50-TCTGGGCTT used. Results were scored by quantitative analysis performed CACGTTCCATA-30 (forward-prA) and 50-GTTCCTTTCT with a computerized analyzer system (Ibas, 2000, Kontron, TGGCCGATGT-30 (reverse-prA) and 50-ATCGGCCAAGA Zeiss) by using as cut-off the staining of X10% of cells AAGGAACCA-30 (forward-prB) and 50-AGCATTCGGCGT displaying evident nuclear signal. TAGACACA-30 (reverse-prB). As specificity control of the binding the primers used were GAPDH 50-GTATTCCCC Statistical analysis CAGGTTTACATG-30 (forward) and GAPDH 50-TTCTCCA For the comparison of statistical significance between two TGGTGGTGAAGAC-30 (reverse). groups of experiments, Student’s t test was used. A P-value of o0.05 was considered statistically significant. Methylation analysis For immunohistochemical studies, differences between the Sodium bisulphite conversion of genomic DNA (about 2 mg for different diagnostic groups were analysed by the nonpara- each conversion) was obtained using Epitect Bisulphite kit metric Kruskal–Wallis one-way ANOVA test. A P-value (Qiagen, Valencia, CA, USA) following the purchaser’s o0.001 was considered statistically significant. instructions. Amplicons used for methylation analysis were obtained from approximately 50 ng of bisulfite treated genomic DNA. Quantitative DNA methylation analysis was performed Databases using the PSQ 96MA instrument from Pyrosequencing The Hand1 gene sequences were retrieved by the Ensembl (Biotage AB, Uppsala, Sweden) following the protocol database. Accession numbers were: Human ENSG000001 suggested by the manufacturer. The reactions were assayed 13196; Mouse ENSMUS00000037335 (transcript ENSMU on the PSQ 96MA using the SNP analysis software provided ST00000036917); Rat ENSRNOG00000002582. by the manufacturer. Primers used for PCR reactions were: B- Hand1FW (-175/-151) 50-aagttygtagagtagggagttgag-30 (50-Bio- Acknowledgements tinylated); Hand1RV (À19/ þ 4) 50-accrcractttaatatcaacctc-30. Amplification were carried out on 10 ng of bisulfite treated This work was supported by grants from the Associazione DNA using HotStarTaq DNA polymerase (Qiagen) under the Italiana Ricerca sul Cancro (AIRC), and the Ministero following conditions: 15 min at 95 1C, followed by 50 cycles of dell’Universita` e della Ricerca Scientifica e Tecnologica 30 s at 95 1C, 40 s at 57,5 1Cand 1 min at 72 1C, then a final (MIUR). This work was supported from NOGEC-Naples elongation of 10 min at 72 1Cbefore holding at 4 1C, in a final Oncogenomic Center. We thank the Associazione Partenopea reaction volume of 50 ml. Sequencing primer (Hand1S1) was per le Ricerche Oncologiche (APRO) for its support. We thank 50-ataactatcccaaattttac-30. The analysis of a non-CpG cytosine Gabriella Sole, INN-CNR of Cagliari, for statistical analysis.

References

Baldassarre G, Fedele M, Battista S. (2001). Onset of natural killer cell Chiappetta G, Avantaggiato V, Visconti R, Fedele M, Battista S, lymphomas in transgenic mice carrying a truncated HMGI-Cgene Trapasso F et al. (1996). High level expression of the HMGI (Y) by the chronic stimulation of the IL-2 and IL-15 pathway. Proc Natl gene during embryonic development. Oncogene 5: 2439–2446. Acad Sci USA 98: 7970–7975. Cross JC, Flannery ML, Blanar MA, Steingrimsson E, Jenkins NA, Battista S, Pentimalli F, Baldassarre G, Fedele M, Fidanza V, Copeland NG et al. (1995). Hxt encodes a basic helix-loop-helix Croce CM et al. (2003). Loss of Hmga1 gene function affects transcription factor that regulates trophoblast cell development. embryonic stem cell lympho-hematopoietic differentiation. FASEB J Development 121: 2513–2523. 17: 1496–1498. Cserjesi P, Brown D, Lyons GE, Olson EN. (1995). Expression of the Berlingieri MT, Pierantoni GM, Giancotti V, Santoro M, Fusco A. novel basic helix-loop-helix gene eHAND in neural crest derivatives (2002). Thyroid cell transformation requires the expression of the and extraembryonic membranes during mouse development. Dev HMGA1 proteins. Oncogene 21: 2971–2980. Biol 170: 664–678.

Oncogene HAND1 regulation by HMGA1 proteins J Martinez Hoyos et al 885 Donato G, Martinez Hoyos J, Amorosi A, Maltese L, Lavano A, Kondo T, Ezzat S, Asa S. (2006). Pathogenetic mechanisms in Volpentesta G et al. (2004). High mobility group A1 expression thyroid follicular-cell neoplasia. Nature Rev Cancer 6: correlates with the histological grade of human glial tumors. Oncol 292–306. Rep 6: 1209–1213. Martinez Hoyos J, Fedele M, Battista S, Pentimalli F, Kruhoffer M, Fedele M, Battista S, Manfioletti G, Croce CM, Giancotti V, Fusco A. Arra C et al. (2004). Identification of the genes up- and down- (2001). Role of the high mobility group A proteins in human regulated by the high mobility group A1 (HMGA1) proteins: tissue lipomas. Carcinogenesis 22: 1583–1591. specificity of the HMGA1-dependent gene regulation. Cancer Res Fedele M, Fidanza V, Battista S, Pentimalli F, Klein-Szanto AJ, 64: 5728–5735. Visone R et al. (2006). Haploinsufficiency of the Hmga1 gene causes Pallante P, Berlingieri MT, Troncone G. (2005). UbcH10 over- cardiac hypertrophy and myelo-lymphoproliferative disorders in expression may represent a marker of anaplastic thyroid carcino- mice. Cancer Res 66: 2536–2543. mas. Br J Cancer 93: 464–471. Fusco A, Fedele M. (2007). Roles of the HMGA proteins in cancer. Reeves R. (2001). Molecular biology of HMGA proteins: hubs of Nat Rev Cancer 7: 899–910. nuclear function. Gene 277: 63–81. Hagihara A, Miyamoto K, Furuta J, Hiraoka N, Wakazono K, Seki S Riley P, Anson-Cartwright L, Cross JC. (1998). The Hand1 bHLH et al. (2004). Identification of 27 50 CpG islands aberrantly transcription factor is essential for placentation and cardiac methylated and 13 genes silenced in human pancreatic cancers. morphogenesis. Nat Genet 18: 271–275. Oncogene 23: 8705–8710. Ritter O, Haase H, Schulte HD, Lange PE, Morano I. (1999). Kaneda A, Kaminishi M, Yanagihara K, Sugimura T, Ushijima T. Remodeling of the hypertrophied human myocardium by cardiac (2002). Identification of silencing of nine genes in human gastric bHLH transcription factors. J Cell Biochem 74: 551–561. cancers. Cancer Res 62: 6645–6650. Takada T, Yagi Y, Maekita T, Imura M, Nakagawa S, Tsao SW Knofler M, Meinhardt G, Bauer S, Loregger T, Vasicek R, et al. (2004). Methylation-associated silencing of the Wnt Bloor DJ et al. (2002). Human Hand1 basic helix-loop-helix antagonist SFRP1 gene in human ovarian cancers. Cancer Sci 95: (bHLH) protein: extra-embryonic expression pattern, interaction 741–744. partners and identification of its transcriptional repressor domains. Thanos D, Maniatis T. (1995). Virus induction of human IFN b gene Biochem J 361: 641–651. expression requires the assembly of an enhanceosome. Cell 83: 1091–1100.

Oncogene